Exploring the role of Profilin 1 in axon formation, growth and regeneration

(1)

Exploring the role

of Profilin 1 in axon

formation, growth

and regeneration

Joana Beatriz Antunes Moreira Carvalho

Marques

Dissertação de Mestrado apresentada à

Faculdade de Ciências da Universidade do Porto e ao Instituto de

Ciências Biomédicas Abel Salazar em

Bioquímica

(2)

of Profilin 1 in axon

formation, growth

and regeneration

Joana Beatriz Antunes Moreira Carvalho

Marques

Mestrado em Bioquímica

Departamento de Química e Bioquímica 2016

Supervisor

Dr Mónica Sousa

Principal Investigator, IBMC/i3S

Affiliate Professor, ICBAS

(3)

(4)

O Presidente do Júri,

(5)

(6)

The greatest obstacle to discovery is not ignorance - it is the illusion of knowledge.

(7)

(8)

Agradecimentos

Em primeiro lugar, gostaria de agradecer à Doutora Mónica Sousa, minha orientadora, por me ter aceite no seu grupo de investigação e pela sua dedicação, exigência e profissionalismo que, sem dúvida, guardarei como um exemplo a seguir e me inspiram a fazer sempre mais e melhor. O meu muito obrigada, também, por todo o seu apoio, disponibilidade e partilha de conhecimento, cruciais para a concretização deste projecto.

Devo, também, um agradecimento especial à Rita Costa, sem a qual esta tese não seria possível, pela paciência em me ensinar e por toda a partilha de experiência e conhecimento. Agradeço a todos os membros dos Nerve Regeneration (não podendo deixar de destacar o Tiago, Fernando, Rita Malheiro, Diogo, Joana, Marina e Jéssica) por me terem acolhido tão bem no grupo, por todo o apoio, dicas e ajuda que me deram sempre que precisei e pela boa disposição contagiante, essenciais ao longo deste ano e que tornaram todo o trabalho mais fácil. Foi um prazer e uma honra poder fazer parte deste grupo e aprender e trabalhar com pessoas tão empenhadas e talentosas.

Agradeço aos meus amigos, tanto os que a vida académica me presenteou como os mais “antigos”, pela companhia, apoio, palavras de encorajamento e, acima de tudo, por tantos bons momentos partilhados.

Não posso deixar de agradecer às pessoas que me acompanham, ajudam e apoiam em tudo há mais de duas décadas de existência e às quais devo tudo aquilo que sou hoje. Aos meus avós maternos, pais e irmã, o meu mais profundo agradecimento por me guiarem ao longo deste difícil caminho que se chama Vida e por me ajudarem a ultrapassar todos os obstáculos. Todas as palavras são poucas para expressar a minha gratidão para convosco.

Por fim, um agradecimento especial ao meu avô Zeca, que sempre olhou e continuará a olhar por mim, onde quer que esteja. Espero deixar-te sempre orgulhoso.

(9)

(10)

Abstract

In the mammalian adult CNS, axons present a very low regenerative capacity after injury, in contrast to PNS neurons. Despite that the cytoskeleton is known to play a crucial role in axon growth and regeneration, the underlying mechanisms are still not fully understood. Actin is extremely important in these processes, as well as the various actin-binding proteins with which it interacts. One of these proteins is profilin 1 (Pfn1), which promotes the exchange of ADP for ATP in actin monomers, increasing the rate of polymerization of the actin filament. Previous results obtained in our group demonstrate that neurons with a high intrinsic regenerative capacity have increased activity of Pfn1, and that acute deletion of Pfn1 precluded axon formation in hippocampal neurons and impaired neurite outgrowth in DRG neurons. Collectively, these results point to an important role of this actin-binding protein in axon regeneration and neuritogenesis. Therefore, in this project we aimed to explore the role of Pfn1 in axon formation, growth and regeneration.

Knowing that axon growth is decreased when Pfn1 is absent, we tested whether Pfn1 overexpression would result in increased axon growth in both hippocampal and dorsal root ganglion (DRG) neurons. Two Pfn1 forms were overexpressed, wild-type (WT) Pfn1 and a phosphorylation-resistant mutated form, which is constitutively active (S137A), and subsequently we evaluated neurite outgrowth, actin and microtubule (MT) dynamics. Our results showed that, in vitro, the overexpression of either Pfn1 WT or Pfn1 S137A, which cannot be negatively regulated by phosphorylation at serine 137, consistently increased neurite outgrowth, actin dynamics and microtubule growth speed. Interestingly, the effect of Pfn1 S137A was generally stronger than that of Pfn1 WT. Taking into account the robust effect of Pfn1 S137A in vitro, we will test the effectiveness of this Pfn1 mutant in enhancing axon regeneration in vivo. This will be achieved by using a viral approach to deliver Pfn1 S137A to DRG neurons of rats with a sciatic nerve injury, followed by evaluation of axon regeneration in the sciatic nerve. If Pfn1 S137A increases axon regeneration in vivo, Pfn1 may become a good target for future therapies to improve axon regeneration following nerve injury.

Moreover, previous data from our group has shown that DRG neurons from mice lacking Pfn1 in neurons (Thy1cre+/-_Pfn1fl/fl_{- Pfn1 cKO) have impaired neurite outgrowth,}

decreased actin retrograde flow and impaired MT dynamics. To determine if there is a compensatory effect following Pfn1 deletion in vivo, the levels of Pfn2 in the brain of Pfn1 cKO mice were evaluated. In the brains of Pfn1 cKO mice, no statistical difference was observed in Pfn2 levels when comparing to WT mice. To further determine possible

(11)

compensatory mechanisms, we assessed neurite outgrowth after the acute deletion of Pfn2 (by shRNA) in Pfn1 cKO DRG neurons and found that the absence of Pfn2 had no effect on neurite growth, whether Pfn1 was absent or not.

Besides profilin role as an actin-binding protein, it may also interact with phosphatidylinositol lipids, in particular PIP2, which is involved in the PI3K/AKT pathway.

This fundamental pathway is involved in several cellular processes, such as MT dynamics. To further understand the molecular role of Pfn1 in this pathway and how it modulates MT dynamics, Pfn1 WT and Pfn1 S137A were overexpressed in a neuronal cell line and the expression levels of some players of the PI3K/AKT pathway were quantified. Pfn1 S137A overexpression led to increased levels of activated AKT and inactivated GSK3β, which suggests that Pfn1 not only affects the actin cytoskeleton, but may also regulate MT dynamics by interfering with an intracellular signalling pathway.

In summary, this work provides important evidences supporting the pivotal role of Pfn1 in axon formation, growth and regeneration.

Keywords: actin dynamics; profilin; axon formation; axon growth; axon regeneration; spinal cord injury; microtubule dynamics.

(12)

Resumo

No sistema nervoso central adulto dos mamíferos, os axónios apresentam uma capacidade regenerativa diminuída após uma lesão, contrariamente aos neurónios do sistema nervoso periférico. Apesar de o citoesqueleto ter um papel crucial no crescimento e regeneração axonal, os mecanismos subjacentes ainda não são totalmente compreendidos. A actina é extremamente importante nestes processos, bem como as diversas proteínas de ligação à actina com as quais esta interage. Uma destas proteínas de ligação à actina é a profilina 1 (Pfn1), que promove a permuta de ADP por ATP nos monómeros de actina, aumentando a taxa de polimerização dos filamentos de actina. Resultados preliminares provenientes do nosso grupo de investigação demonstraram que neurónios, cuja capacidade intrínseca de crescimento se encontra aumentada, exibem um aumento da atividade de Pfn1. Por outro lado, a deleção aguda desta proteína resultou numa diminuição do crescimento de neurites, indicando que a Pfn1 tem um papel importante no contexto de regeneração axonal. Assim, neste projeto, o principal objetivo consistiu em investigar o papel da Pfn1 na formação, crescimento e regeneração axonal.

Dado que o crescimento axonal diminui na ausência de Pfn1, foi testado se a sua sobreexpressão resultaria num aumento do crescimento de neurites, tanto em neurónios de hipocampo como em neurónios dos gânglios da raiz dorsal (GRD). Para tal, foi efetuada a sobreexpressão de duas formas de Pfn1, a forma selvagem (Pfn1 WT) e uma forma mutada fosforesistente, constitutivamente ativa (Pfn1 S137A), e, posteriormente, avaliou-se o crescimento de neurites, bem como a dinâmica do citoesqueleto de actina e microtúbulos. Os resultados obtidos demonstraram que, in vitro, a sobreexpressão de qualquer uma das duas formas de Pfn1 levou a um aumento consistente do crescimento de neurites, bem como da dinâmica de actina e dinâmica de microtúbulos. Curiosamente, o efeito da Pfn1 S137A foi, em geral, mais pronunciado do que o da Pfn1 WT. Tendo em conta o efeito robusto da Pfn1 S137A in vitro, iremos testar a eficiência deste mutante em promover a regeneração axonal in vivo. Para tal, será utilizada uma abordagem viral para sobreexpressar a Pfn1 S137A in vivo, em neurónios de GRD de ratazanas que sofreram uma lesão no nervo ciático, seguida de uma avaliação da regeneração axonal no nervo ciático. Se a Pfn1 S137A levar a um aumento da regeneração axonal in vivo, a Pfn1 poderá constituir um bom alvo em futuras terapias para promover a regeneração após uma lesão de um nervo.

Além disso, resultados preliminares obtidos no nosso grupo demonstraram que neurónios de GRD de murganhos que não possuem Pfn1 (Thy1cre+/-_Pfn1fl/fl_{- Pfn1 cKO)}

(13)

apresentavam uma redução no crescimento neuronal, bem como diminuição da dinâmica do citoesqueleto de actina e microtúbulos. Por forma a determinar se existe um efeito compensatório da Pfn2 nos neurónios acima mencionados, os níveis de Pfn2 foram avaliados em amostras de cérebro de murganhos Pfn1 cKO. Verificou-se que, em murganhos Pfn1 cKO, não há diferença estatisticamente significativa nos níveis de Pfn2 quando comparado com murganhos WT. Adicionalmente, também se avaliou o crescimento de neurites após a deleção aguda de Pfn2 (através de shRNA) em neurónios de GRD de murganhos Pfn1 cKO e observou-se que a deleção de Pfn2 não tem qualquer efeito sobre o crescimento de neurites, tanto na presença como ausência de Pfn1.

Além do papel da profilina como proteína de ligação à actina, esta proteína também pode interagir com lípidos fosfatidilinositol, em particular com PIP2, que está

envolvido na via de sinalização PI3K/AKT. Esta via está envolvida em diversos eventos celulares, tal como na dinâmica de microtúbulos. Para se tentar perceber melhor o papel da Pfn1 nesta via e como poderá influenciar a dinâmica de microtúbulos, a Pfn1 WT e a Pfn1 S137A foram sobreexpressas numa linha celular neuronal e os níveis de expressão de algumas moléculas intervenientes na via PI3K/AKT foram avaliados. A sobreexpressão da Pfn1 S137A conduziu a um aumento dos níveis da forma ativada da proteína AKT e da forma inativada da proteína GSK3β, o que sugere que a Pfn1 não só afeta o citoesqueleto de actina, como também pode regular a dinâmica de microtúbulos através de uma via de sinalização intracelular.

Em suma, este trabalho fornece importantes evidências que suportam o papel essencial da Pfn1 na formação, crescimento e regeneração axonal.

Palavras-chave: dinâmica de actina; profilina; formação axonal; crescimento axonal; regeneração axonal; lesão da espinal medula; dinâmica de microtúbulos.

(14)

List of Figures

Figure 1 – Cytoskeleton components. ... 2

Figure 2 – Stages of neuronal polarization in vitro... 5

Figure 3 – The growth cone structure... 8

Figure 4 – An overview of some actin-binding proteins and their functions. ... 10

Figure 5 – Intrinsic mechanisms of axonal regeneration in PNS neurons after injury. . 16

Figure 6 – Profilin structure and its binding domains. ... 18

Figure 7 – Pfn1 activity increases after a conditioning lesion. ... 23

Figure 8 – Pfn1 is a regulator of neuritogenesis and axon growth. ... 24

Figure 9 – Depletion of Pfn1 impairs actin and microtubule dynamics in the growth cones of DRG neurons. ... 25

Figure 10 – Overexpression of Pfn1 S137A leads to increased neurite outgrowth of hippocampal neurons. ... 37

Figure 11 – Overexpression of Pfn1 WT and Pfn1 S137A leads to increased neurite outgrowth of DRG neurons. ... 38

Figure 12 – Overexpression of Pfn1 WT and Pfn1 S137A increases actin and MT dynamics in hippocampal neurons. ... 39

Figure 13 – Overexpression of Pfn1 WT and Pfn1 S137A increases actin and MT dynamics in DRG neurons. ... 39

Figure 14 – Vector map designed for the in vivo delivery of Pfn1 S137A. ... 40

Figure 15 – Restriction analysis of pAAV-hPfn1S137A-IRES-GFP with either SmaI or HindII. ... 41

Figure 16 – DRG neurons show high expression levels of GFP after transfection with the AAV-based construct. ... 41

Figure 17 – Profilin 2 expression in brain from WT and Pfn1 cKO mice. ... 42

Figure 18 – Pfn2 acute deletion in DRG neurons from WT and Pfn1 cKO mice does not influence neurite outgrowth. ... 43

Figure 19 – Pfn1 may interfere with PI3K/AKT signalling pathway. ... 45

Figure 20 – Pfn1 may interfere with signalling cascades and function as bridge between actin and microtubules. ... 45

(17)

(18)

List of Abbreviations

AAV Adeno-Associated Virus ABP Actin-Binding Protein

ADF Actin-Depolymerizing Factor ADP Adenosine Diphosphate AKT Protein Kinase B

ALS Amyotrophic Lateral Sclerosis APC Adenomatous Polyposis Coli protein Arg1 Arginase 1

Arp2/3 Actin-Related Protein 2 and 3 complex ATF3 Activating Transcription Factor 3 ATP Adenosine Triphosphate

BBB Blood–Brain Barrier BSA Bovine Serum Albumin CAD Cath.-a-differentiated

cAMP cyclic Adenosine Monophosphate CAP Cyclase-Associated Protein CGN Cerebellar Granule Neurons cKO Conditional Knockout CL Conditioning Lesion CMV Cytomegalovirus

CNS Central Nervous System

CRMP2 Collapsin Response Mediator Protein 2 CSPG Chondroitin Sulphate Proteoglycan DAPI 4',6-Diamidino-2-Phenylindole DIV Day(s) in vitro

DLK-1 Dual Leucine Kinase 1

DMEM Dulbecco's Modified Eagle Medium DNA Deoxyribonucleic acid

DRG Dorsal Root Ganglia

DsRed Discosoma sp. red fluorescent protein

E Embryonic day(s)

EB3 End-Binding protein 3 ECM Extracellular Matrix

(19)

EDTA Ethylenediaminetetraacetic acid

EGFP Enhanced GFP

Elk-1 ETS domain-containing protein

ERK Extracellular signal-Regulated Kinase ERM Ezrin–Radixin–Moesin

F-actin Actin filament FBS Fetal Bovine Serum G-actin Globular actin

GAP-43 Growth-Associated Protein-43 GDP Guanosine Diphosphate GFP Green Fluorescent Protein GSK3 Glycogen Synthase Kinase 3 GTP Guanosine Triphosphate HBSS Hanks' Balanced Salt Solution HDAC5 Histone Deacetylase 5

HPRT Hypoxanthine-guanine Phosphoribosyltransferase HRP Horseradish Peroxidase

IF Intermediate Filaments

Ig Immunoglobulin

IL-6 Interleukin-6

IRES Internal Ribosome Entry Site ITR Inverted Terminal Repeat JNK c-Jun N-terminal Kinases

KD Knockdown

LIMK LIM domain Kinase

MAG Myelin-Associated Glycoprotein MAP Microtubule-Associated Protein

MT Microtubules

mTor mammalian Target of rapamycin N-cadherin Neural cadherin

NGF Nerve Growth Factor NMDA N-methyl-D-aspartate

NMRI Naval Medical Research Institute

NPY Neuropeptide Y

OMgp Oligodendrocyte Myelin glycoprotein ori origin of replication

(20)

PAR1 Protease-Activated Receptor 1 PBS Phosphate-Buffered Saline PCR Polymerase Chain Reaction

PDK1 3-Phosphoinositide Dependent protein Kinase-1 PenStrep Penicillin-Streptomycin

PFA Paraformaldehyde

Pfn Profilin

PI Phosphatidylinositol PI3K Phosphoinositide 3-Kinase PIP2 PI-(4,5)-bisphosphate

PKA Protein Kinase A PKC Protein Kinase C

PLC Phospholipase C

PLL Poly-L-Lysine PLP Poly-L-Proline

PMGS Plasma Membrane Ganglioside Sialidase PNS Peripheral Nervous System

PTEN Phosphatase and Tensin Homologue Rac1 Ras-related C3 botulinum toxin substrate 1 RAG Regeneration-Associated Gene

RhoA Ras homolog family member A RNA Ribonucleic Acid

ROCK Rho-associated protein kinase RPMI Roswell Park Memorial Institute

RT Room Temperature

SCI Spinal Cord Injury

SDS-PAGE Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis SEM Standard Error of the Mean

shRNA small hairpin RNA

SLICK Single-neuron Labeling with Inducible Cre-mediated Knockout SMA Spinal Muscular Atrophy

SMN Survival of Motor Neuron

snRNP small nuclear Ribonucleoprotein

STAT3 Signal Transducer and Activator of Transcription 3 TBS Tris Buffered Saline

(21)

VIP Vasointestinal Peptide

WASP Wiskott–Aldrich Syndrome Protein

WAVE WASP-family Verprolin-homologous protein

WT Wild-Type

(22)

Introduction

Neurons are probably one of the most striking examples of polarization, having two distinct compartments: the somatodendritic compartment - comprising the cell body (soma) and the dendrites - and the axonal compartment. Neurons rely on this highly polarized shape to perform their functions that depend on dendrites to receive information that will be integrated in the cell body and then transmitted through the axon. Dysfunction in any of these mechanisms has been associated with neurodegenerative diseases that, nowadays, increasingly affect the human population given the increased life expectancy. Thus, having a deeper insight into the intracellular machinery and cellular events involved in the establishment and maintenance of neuronal polarity may be part of the key to thrive in discovering new therapeutics for neurodegenerative disorders.

1. Neuronal cytoskeleton and polarization

The cytoskeleton consists of a dense and highly dynamic network formed by three distinct classes of interacting structural complexes, with quite different properties, namely: microtubules (MTs), intermediate filaments (IFs) and microfilaments (or actin filaments) (Figure 1). Each of these elements exists concurrently and interdependently in overlapping cellular domains and each one presents a characteristic composition, structure and organization. [1] The cytoskeleton is involved in a plethora of critical functions across the cell, from determining its shape and architecture to being involved in intracellular transport (such as organelle trafficking), but also plays a role in neuronal growth and secretion, cell motility, interaction with the extracellular environment or in the formation of presynaptic and postsynaptic specializations, among other cellular events.[2] Despite the word ‘skeleton’, cytoskeleton is not a stiff or static structure, but rather a dynamic and adaptive one, which can be arranged into bundles or networks. Due to its adaptability and tight regulation, it allows cells to sense external mechanical and/or chemical signals and respond accordingly, functioning as a link between the intracellular and extracellular environments. [3]

As aforementioned, the cytoskeleton’s main components although structurally different interact with one another in a variety of ways. Briefly, microtubules, which are the stiffest of the three polymers, consist of hollow tube-like structures (24 nm in diameter), whose walls are typically formed by 13 protofilaments. [4] In turn, these

(23)

protofilaments are composed of α- and β-tubulin heterodimers that align in a head-to-tail fashion and possess GTPase activity. Tubulin dimers can bind two molecules of GTP, one of which can be hydrolysed to GDP subsequent to incorporation in the microtubule filament. This allows MTs to switch between growing and shrinking phases and this ‘dynamic instability’ enables both individual microtubules to search the cellular space quickly and the microtubule cytoskeleton to reorganize rapidly. [2, 5] Actin filaments (F-actin) are especially enriched in certain subcellular neuronal regions, such as the growth cone or dendritic spines and consist of two strands of polymerized globular actin monomers (G-actin) arranged in a helix. Actin polymerization is tightly regulated by actin-binding proteins (ABPs) and occurs preferentially at the fast growing end (or ‘barbed end’), while depolymerization occurs in the opposite end of the filament (the ‘pointed end’). [6] Unlike actin filaments and microtubules, which are assembled from one or two proteins, intermediate filaments (or neurofilaments, in neurons) may be assembled from a large diverse family of proteins. Despite this, intermediate filaments are structurally similar, regardless of the type, forming 8–10 nm rope-like filaments that may be several micrometres long. Furthermore, IFs are the less rigid of the three cytoskeleton components and, as they are not polarized structures (unlike actin and MTs), IFs cannot support directional movement of molecular motors. [6, 7]

Figure 1 – Cytoskeleton components. Neurons have a cytoskeleton that consists of three main polymers: microtubules

(green), neurofilaments (purple) and actin filaments (red). [3]

Because of the central role of cytoskeleton in cell structure and in many cellular events, as previously mentioned, even a slight alteration in a cytoskeleton component or

(24)

binding protein might have a profound impact in cells, which can eventually lead to marked pathological conditions. There are some reports in the literature of cytoskeleton alterations which can lead to a variety of phenotypes, such as: tumorigenic phenotypes, in which the regulation of the actin cytoskeleton is unbalanced [8, 9], myopathies, skin and epithelial diseases and also neurodegenerative disorders, which are typically characterized by abundant abnormal aggregates of cytoskeletal proteins. [10, 11]

Furthermore, the cytoskeleton plays a crucial role in the establishment and maintenance of cell polarity, which is essential in many cell types, but especially important in neurons, in order to allow its correct functioning. [12, 13] Interestingly, two of the three cytoskeleton components (namely actin and MTs) present an intrinsically polarized structure, as previously referred, which results in distinct polymerization and depolymerization rates in each end. This structural difference is crucial for the cytoskeleton to perform its role in neuronal polarization and therefore, the comprehension of how actin and MTs contribute to cell polarity and its regulation is of particular importance.

Of note, during neuronal polarity, neurons also rely on several signalling pathways and respective signal transduction molecules, which are activated during axon specification, formation and elongation and regulate cytoskeleton dynamics, such as phosphoinositide 3-kinase (PI3K), phosphatase and tensin homologue (PTEN) or glycogen synthase kinase 3 beta (GSK3β). [14] Several reports revealed that PI3K is implicated in axon specification, since its pharmacological inhibition prevented axon formation. [14, 15] Briefly, PI3K catalyzes PIP2 phosphorylation, originating PIP3, which

was found to be selectively accumulated within a single neurite of hippocampal neurons at stage 2 of development. [15] On the other hand, PTEN catalyzes the opposite reaction, promoting PIP3 dephosphorylation into PIP2 and consequently limiting PIP3 signalling.

The formation of PIP3 by PI3K recruits AKT to the membrane and promotes its activation,

which once activated, may phosphorylate its downstream targets. [14] Curiously, the active form of AKT has been reported to be enriched in the growth cones of polarized neurons, [16] highlighting the importance of this molecule and the signalling pathways in which it is involved. A common AKT target is GSK3β, which becomes inactive when phosphorylated in serine 9 residue, therefore preventing phosphorylation of its targets, some of which are potential effectors of neuronal polarity and are involved in cytoskeleton regulation (such as adenomatous polyposis Coli protein (APC), collapsin response mediator protein 2 (CRMP2), tau and other microtubule-associated proteins (MAPs)). [17]

(25)

1.1. Stages of neuronal polarization

Polarity enables cells to carry out specialized functions and it is characterized by the asymmetric accumulation of mobile components, such as regulatory molecules, in opposite poles of the cell and by the oriented organization of the inherently polarized cytoskeletal filaments (actin and MTs) along the axis of polarity. [13] Neurons are among the most polarized cell types, which classically present a round cell body and two morphologically and functionally distinct types of processes - axon and dendrites. The axon is a long and thin structure, whereas dendrites are relatively short and tapered, becoming thinner with increased distance from the cell body. [6]

This polarized morphology is fundamental for the transmission of neuronal information, as the dendrites receive synaptic input from other neurons, which is integrated at the soma and propagated along the axon to the presynaptic terminal, where the signal is transmitted to target neurons through synapses. [18, 19] Therefore, polarization is an indispensable event during neuronal development and also for the integration and transmission of information in the entire nervous system.

During polarization, neurons experience different developmental stages, starting from a small, round and symmetric sphere, which will develop into a highly polarized mature neuron. However, the timing of each step may vary greatly among the different neuron types. [18] This process has been extensively studied, predominantly in rat embryonic hippocampal neurons [20, 21] and postnatal cerebellar granule neurons (CGN) [22], which allowed to establish a well-characterized model for the in vitro neuronal polarization process. The morphological events that occur during embryonic hippocampal neurons polarization can be divided in five robust stages (Figure 2).

Shortly after being plated, neurons attach to the substrate as round spheres surrounded by lamellipodia and filopodia, which are actin-rich structures (stage 1). A few hours later, they start forming various minor processes, which are termed immature neurites that are the precursors of the future axon and dendrites (stage 2). This process is called neuritogenesis and at this stage, the neuron is still similar to an unpolarized, symmetric sphere. These immature neurites are morphologically identical and display repeated, random cycles of extension and retraction. After neuritogenesis, one of the neurites, which will develop into the future axon, begins to extend rapidly, becoming much longer than the other neurites (stage 3). [20] Simultaneously, the growth of the remaining neurites is inhibited and they continue to undergo alternating spurts of growth and retraction, maintaining their length until they develop into dendrites, at later stages (stage 4). Finally, the maturation is achieved by the formation of dendritic spines and the

(26)

establishment of synapses, which will enable the transmission of signals between neurons (stage 5). [23, 24]

Figure 2 – Stages of neuronal polarization in vitro. Representation of morphological changes that occur during

neuronal polarization. Right after plating, neurons adhere to the substrate and exhibit intense lamellipodial and filopodial protrusive activity (Stage 1), which eventually leads to the formation of multiple immature neurites (Stage 2). Then, there is a break of the neuronal symmetry when one of the neurites starts to grow more rapidly than the others, originating the axon (in red) (Stage 3), while the remaining neurites will become dendrites. At stage 4, axon and dendritic growth occur, as well as branching and subsequently, after several days in culture, neurons develop dendritic spines and synapses are established (stage 5). (Below) Phase-contrast images of hippocampal neurons during the first 13 d of culture. Scale bars: 25 µm. Adapted from [23] and [25].

It is important to mention that this in vitro model of polarization is based on cultures of embryonic rat hippocampal neurons, in which most plated cells are polarized postmitotic neurons prior to dissociation that may preserve some aspects of the original polarization. For this reason, neuronal polarization using this in vitro model likely corresponds to the repolarization of previously polarized neurons in vivo and does not reproduce exactly the conditions that neurons face during development. [23] Notwithstanding, hippocampal neuron cultures are still extremely useful to understand the cell biology events that lead to polarization and, undoubtedly, the formation and elongation of the axon is the crucial event in breaking cell symmetry and establishing neuronal polarity.

1.2. Cytoskeleton dynamics during axon growth and neuronal polarization

It is known that between stages 2 and 3 of polarization, symmetry is broken when one of the immature neurites starts to grow faster and becomes the axon, whereas the

(27)

growth of other processes is inhibited, as previously mentioned. Even though external stimuli play a role in the establishment of neuronal polarity, neurons definitely possess intrinsic mechanisms that govern symmetry breaking and allow the maintenance of that morphology over time. [18] Those intrinsic mechanisms greatly depend on actin and MTs dynamics. [26]

In fact, it was observed in cultured hippocampal neurons that the axon formation is preceded by an enlargement and increased dynamics of the growth cone in one of the neurites, showing an unstable and loose actin meshwork. [27] The increased actin turnover in one of the neurites is enough to confer axonal identity and as it is less dense, allows MTs to more easily and rapidly protrude into distal regions of the axonal growth cone, sustaining axonal extension. [28] Contrary to the actin cytoskeleton, MTs stabilize in that particular neurite before axon formation and become even more stable in the future axon shaft of the polarized neuron. [29] Consequently, that specific neurite will elongate rapidly, originating an axon that grows at a rate between five to ten times faster than the other neurites. [20]

On the other hand, the other neurites remain small and in a state of undynamic quiescence, presenting a denser actin meshwork. This creates a non-permissive environment that sterically hinders microtubule advance, characterized by stable and condensed actin filaments. Therefore, MTs are not able to protrude and stabilize in those neurites, which will eventually develop into dendrites at later stages. [28]

Even though the above reported evidences clearly demonstrate that actin and MTs work together in the establishment of polarity, the underlying mechanisms that trigger initial actin dynamics and MTs stabilization are still not fully understood. However, a variety of factors have already been implicated in MTs stabilization in axons, namely

GSK3β, protease-activated receptor 1 (PAR1), CRMP2, tau and MAP1b.Nevertheless,

it is still unclear which of these factors are involved in the initial stabilization of MTs in

the axon and how they might be recruited to the presumptive axon. [13, 18]Interestingly,

it was observed that when the actin depolymerizing reagents Cytochalasin D or Latrunculin B were added to neuronal cultures, neurons had multiple axon-like processes, emphasizing the importance of the destabilization of the actin cytoskeleton in axon formation and extension. [27]

Furthermore, it has also been reported that even after the axon is specified, other neurites have the potential to change their identity during development. It was observed

that the transection of axons of hippocampal neurons early in development [21] or even of mature neurons [30], could cause the polarity to change and thus a minor neurite

(28)

transection of the axon, if one of the neurites was at least 10 μm longer than the others,

it would invariably begin to grow and become the axon. However, in mature neurons this

change in polarity depended on the length of the cut axon. If the axon was cut at a

distance inferior to 35 μm from the cell body, polarity would change and another dendrite would become the axon; if the cut was further than 35 μm, the axon would regenerate. Therefore, these observations demonstrate that the neuronal polarization is a plastic process, not only early in development, but also at later stages, when neurons are integrated in functional networks.

1.3. The Growth Cone

As previously stated, the cytoskeleton plays a crucial role in the establishment of neuronal polarity, in particular in the intense rearrangements it undergoes at the tip of the neurite, which is also denominated growth cone (Figure 3). The growth cone is a highly dynamic and specialized structure, which is responsible for the axon growth and allows neurons to sense and explore the extracellular environment, in response to multiple cues. Regarding axon growth in neurons, the growth cone plays an essential role in controlling both the rate and the direction of growth. [31]

Growth cones may assume different shapes and sizes and they constantly extend and retract membrane protrusions, in order to sense the extracellular environment. The structure of the growth cone is fundamental for it to perform its function and it is composed of three main domains that can be distinguished by their characteristic cytoskeletal organization, namely: the central (C) domain, the transition (T) zone and the peripheral (P) domain. [32]

The P domain consists of dynamic finger-like protrusions (termed filopodia) that explore the surrounding environment, separated by sheets of membrane between the filopodia, which are called lamellipodia. Filopodia are formed by long, bundled actin filaments (F-actin bundles), whereas flat mesh-like branched F-actin networks confer support and structure to lamellipodia. In addition, individual dynamic microtubules explore the P domain, usually along the F-actin bundles of filopodia, which may act as guidance sensors. In fact, filopodia may also have a major role in establishing growth cone–substrate adhesive contacts during environmental exploration, by functioning as points of attachment to the substrate and therefore producing tension that can be used for growth cone progression. [33] In turn, the C domain comprises stable, bundled MTs that enter the growth cone from the axon shaft, which support the constant shuttling of organelles and vesicles. Lastly, the T zone lies between the C and P domains. It is

(29)

enriched in actomyosin contractile structures (actin arcs), which are oriented perpendicular to the F-actin bundles, forming a hemicircumferential ring that in a certain extent hinders MTs protrusion to the P domain. [32, 34]

Figure 3 – The growth cone structure. The cytoskeletal elements in the growth cone underlie its shape, and the growth

cone can be separated into three distinct domains based on cytoskeletal distribution: central (C) domain, transition (T) zone and peripheral (P) domain. The P domain consists of long F-actin bundles (filopodia) separated by flattened mesh-like branched F-actin networks (lamelipodia). Additionally, individual dynamic microtubules explore this region, along the F-actin bundles. The C domain comprises stable, bundled microtubules that enter the growth cone from the axon shaft, allowing the transport of numerous organelles and vesicles. Regarding the T zone, it lies in the interface between the C and P domains, with actomyosin contractile structures (actin arcs) perpendicular to F-actin bundles, forming a hemicircunferential ring. Adapted from [32]

Regarding cytoskeleton’s orientation in the growth cone, in particular actin cytoskeleton, it is known that the barbed ends of actin filaments of both filopodia and lamellipodia are oriented towards the leading edge of the growth cone, where incorporation of G-actin monomers in F-actin occurs, whilst the pointed ends face the T zone, where depolymerization takes place. The continuous addition and disassembly of actin monomers (a process called actin treadmilling) ensures that the polymer maintains a constant length and at the same time promotes protrusion of the leading edge membrane and growth cone motility. These dynamic properties of actin, namely F-actin treadmilling combined with F-actin retrograde flow (the continuous movement of F-actin from the leading edge towards the centre of the growth cone) lie in the basis of the growth cone movement in response to directional cues. [31] It has been demonstrated that

(30)

F-actin retrograde flow is driven both by the contractility of myosin II and the push force exerted by actin polymerization in the leading membrane of the P domain. [35] Briefly, myosin II compression in the T zone combined with the pushing force from leading edge actin polymerization may cause buckling and severing of F-actin bundles in the proximal ends, which might also involve the actin-depolymerization factor (ADF)/cofilin. Consequently, the actin fragments can be recycled into actin monomers, which become available for the continuous actin polymerization at the leading edge. However, the difference between the rates of actin polymerization/depolymerization and retrograde flow is not enough to determine if the growth cone extends or retracts. In fact, a hypothesis, called the ‘clutch’ hypothesis has been proposed. Authors suggested that when a growth cone receptor binds to an adhesive substrate this leads to the formation of a complex that acts like an ‘anchor’ or molecular clutch, mechanically coupling the receptors and F-actin flow. Therefore, F-actin becomes anchored, preventing retrograde flow and promoting actin-based forward progression of the growth cone. [32]

The growth cone progression process upon the encountering of attractive adhesive substrates can be divided into three main phases: protrusion, engorgement and consolidation. [36] Briefly, as the receptors at the distal end of the growth cone bind to an adhesive substrate, intracellular signalling cascades are activated and so a molecular ‘clutch’ that links the substrate to the actin cytoskeleton is formed. This will counterbalance the actin retrograde flow and as F-actin polymerization continues in front of the clutch site, filopodia and lamellipodia will move forward and extend. Subsequently, engorgement occurs after F-actin bundles between the adhesion site and the C domain disaggregate and the F-actin arcs reorganize from the C domain towards the site of new growth. [37] This will allow the invasion of C domain MTs into the protruding area, also supporting the transport of organelles and vesicles into this region. Finally, consolidation occurs when the proximal part of the growth cone compacts to form a new segment of the axon shaft. This is promoted by the compression exerted in MTs by myosin II-containing actin arcs, followed by MT-associated protein stabilization. These three stages take place both during the formation of nascent axons or during axon branching.[38]

(31)

1.4. Actin-Binding Proteins (ABPs)

In cells, the assembly and disassembly of actin filaments, as well as their organisation into functional networks, is regulated by a plethora of actin-binding proteins. Therefore, the relative abundance of the different classes of ABPs will dictate the actin architecture at a given place. ABPs participate in a wide range of events involving actin, such as actin polymerization and depolymerization, capping, severing, cross-linking or nucleation of actin filaments (Figure 4). [39]

Figure 4 – An overview of some actin-binding proteins and their functions. Actin-binding proteins (ABPs) interact

with actin in different ways. In the actin treadmilling, cofilin promotes F-actin severing while profilin supports ADP to ATP exchange in G-actin monomers, so that they can be incorporated in free barbed ends. On the other hand, there are other ABPs that promote nucleation, such as Arp2/3 complex (made up of two Arp proteins and five associated subunits) that interacts with the sides of existing actin filaments, forming branched F-actin arrays and formins that capture actin monomers to nucleate a new filament. Additionally, some proteins facilitate the formation of F-actin structures, by promoting cross-linking or bundling, such as filamin, fimbrin or α-actinin. Actin filaments can also become capped by some ABPs, such as CapZ or gelsolin, therefore blocking incorporation of new actin monomers. Adapted from [40]

Actin is an extremely abundant protein in the growth cone, as it can reach a cytoplasmic concentration of up to 100 µM, a value much higher than the 0.1 µM critical

(32)

concentration needed for actin spontaneous polymerization. Interestingly, about half of the actin in the growth cone is in the monomeric form (i.e. unpolymerized), due to the abundance of ABPs that regulate several facets of actin dynamics. [41]

As mentioned before, during axon growth, actin polymerization takes place at the growth cone, pushing the plasma membrane forward. This actin polymerization requires availability of G-actin monomers, which is essentially regulated by two proteins, namely profilin and thymosin β4. Profilin’s main function is to accelerate the ADP to ATP exchange in G-actin monomers released from F-actin pointed ends, thus providing a pool of ATP-actin monomers that can be readily available for polymerization. [42] In turn, thymosin β4 binds and sequesters ATP-G-actin monomers, preventing their incorporation in actin filaments. However, when the free concentration of ATP-G-actin monomers drops, thymosin β4 readily releases them, thus functioning as a ‘buffering system’. As profilin shows a higher affinity for actin than thymosin β4, this allows profilin to maintain a pool of actin monomers available for polymerization, while thymosin β4 holds the rest of the monomers in reserve. [43]

Besides availability of ATP-G-actin monomers, accessible free F-actin barbed ends are also required for actin polymerization. In order to produce new barbed ends, there are three possible mechanisms regulated by different ABPs, namely: nucleation, severing or uncapping of existing filaments. [44] Specifically, actin depolymerizing factor (ADF) and cofilin, which are abundant at the growth cone leading edge, promote actin severing and depolymerization by binding to ADP-F-Actin. Therefore, in combination with profilin, they promote actin treadmilling, by increasing both the number of available free barbed ends and G-actin monomers. [41, 45] There is also another ABP that promotes F-actin severing, named gelsolin, but its role in growth cone dynamics is minor when compared with ADF/cofilin. This protein is able to promote severing of actin filaments in the presence of micromolar calcium and caps the resulting free barbed ends, preventing monomer addition. [46]

There are other ABPs that also regulate the availability of free F-actin barbed ends by capping them, such as CapZ that binds barbed ends and blocks monomer addition, preventing polymerization. However, this capping activity may be inhibited by Ena/VASP (vasodilator stimulated phosphoprotein) proteins. The process of capping may seem counterproductive, since it inhibits F-actin elongation and consequently cell motility or axon growth. However, capping serves as a mechanism to control the length of actin filaments and also avoids non-productive consumption of actin monomers, thus channelling efforts to a limited number of barbed ends. [43]

(33)

Moreover, there are also ABPs that nucleate G-actin to create new barbed ends for polymerization, for example the actin-related protein (Arp) 2/3 complex and formins. The Arp2/3 complex, made up of two Arp proteins and five associated subunits, binds at the side of an existing actin filament and nucleates a new filament that branches from a pre-existing one, supporting the formation of branched actin networks. [47] Formins capture several actin monomers to nucleate a new filament and remain bound to the barbed end to stabilize the nascent filament and facilitate polymerization of long filaments. Both Arp2/3 complex and formins individually bind profilin-ATP-G-actin, enhancing G-actin incorporation. [48]

Additionally, the effective transformation of actin polymerization into protrusion requires that F-actin interacts with membrane components. These interactions can be mediated by ABPs, such as the ezrin–radixin–moesin (ERM) proteins that can bind F-actin to several membrane proteins (for example, L1 - a neuronal adhesion molecule). F-actin can also interact with N-cadherin through α- and ß-catenin [49] or with integrins (which mediate growth cone adhesion to extracellular matrix) via talin, vinculin and α-actinin. These connections between F-actin and adhesive molecules are crucial in order to allow growth cone migration and axon elongation. [43]

2. Regeneration in the Central and Peripheral Nervous System

It is widely known that adult mammalian central nervous system (CNS) neurons present a very restricted regenerative capacity and any attempts to regenerate upon an injury are usually abortive. The regenerative process may be regarded as a recapitulation of neuronal development, since neurons with severed axons must form a new growth cone in order to allow axonal extension and re-establish neuronal connections with their targets. In fact, during development neurons execute a transcription-dependent program of axonal growth that relies in signalling pathways, which allows neurons to successfully perform axon elongation until reaching their post-synaptic targets. However, following this initial phase, the developmental axon growth capacity drops. [50, 51] In contrast to the CNS, after an injury, adult peripheral nervous system (PNS) neurons can reactivate this cell intrinsic program and spontaneously regrow to a significant extent. For that reason, PNS neurons are frequently used as a model to study the process of regeneration and to dissect the underlying mechanisms and players. Interestingly, some studies demonstrated that the removal or neutralization of extracellular inhibitory molecules is not enough to allow successful regeneration of CNS neurons, since they display an incomplete axon regeneration in vivo. [52] Therefore, this highlights the

(34)

importance of both extrinsic factors and cell intrinsic mechanisms in regeneration after injury and the need to better understand them. It is also important to mention that, although PNS regeneration may seem quite remarkable, frequently regenerating axons are misdirected and reinnervate inappropriate targets, leading to incomplete functional recovery. [53]

After an axotomy, neurons present typical morphological changes, referred to as chromatolysis. The elicited acute reactions are analogous in both PNS and CNS neurons, namely: dispersal of the Nissl substance, displacement of the nucleus to the cell’s periphery, swelling of the cell body and loss or retraction of synaptic terminals. Regarding long-term responses, these vary significantly between regeneration-competent (PNS) and inregeneration-competent (CNS) neurons. In the first ones, the cell bodies remain hypertrophic and exhibit signs of protein synthesis and increased metabolism, whereas in CNS neurons cell bodies appear atrophic, with reduced cell volume and dendritic arborization. [52, 54] On the other hand, there are also alterations in the axonal compartment, specifically the distal part of the axon endures Wallerian degeneration and the proximal lesion site reseals the damaged axonal membrane. [55] This is the critical step for regeneration, since the transformation of the severed axonal stump into a competent growth cone that can integrate intra- and extracellular signals is of the most importance, in order to allow growth and extension. [56] Remarkably, whereas peripheral axons can restore a growth cone and start to regrow in the first 24 hours post-injury, CNS axons form a retraction bulb (also called frustrated growth cone). [57] This structure is a hallmark of degenerating axons and typically displays an accumulation of anterogradely transported vesicles and mitochondria and a disorganized network of MTs. [52]

2.1. Extrinsic factors

As previously mentioned, there are substantial differences between the conditions that PNS and CNS neurons encounter upon axonal injury. One such difference is the highly inhibitory glial scar that is formed after CNS injury and which creates a mechanical and chemical barrier to axonal regeneration. The glial scar is particularly rich in reactive astrocytes and proteoglycans, but microglia and oligodendrocytes can also be recruited. [58] Astrocytes can become hypertrophic and produce higher quantities of intermediate filaments and ECM proteins, such as chondroitin sulphate proteoglycans (CSPGs), which are extremely inhibitory to axon outgrowth. Despite the strong detrimental effect of the glial scar, it may also provide important beneficial functions for stabilizing fragile

(35)

CNS tissue after injury, such as promoting repair of the blood–brain barrier (BBB) to prevent an overwhelming inflammatory response and limit cellular degeneration. [59]

In addition, if the myelin structure is damaged during CNS injury, axons also become exposed to myelin-associated inhibitors of regeneration. Several classes of these proteins have already been identified, such as Nogo [60], oligodendrocyte myelin glycoprotein (OMgp) [61] and myelin-associated glycoprotein (MAG) [62], which are all expressed in CNS oligodendrocytes, except MAG that is expressed in PNS Schwann cells as well. Furthermore, other molecules that are upregulated in the core of the lesion and also contribute to the growth retarding effects of the glial scar have been described. For example, semaphorin 3, whose expression increases in fibroblasts that penetrate deeply in the lesion, and ephrin E3, produced by oligodendrocytes.

2.2. Intrinsic Factors

After axonal injury, neurons trigger a complex injury signalling cascade in order to reprogram the neuronal cell body to enter a pro-regenerative program. When the axon is severed, the axon’s interior is exposed to the extracellular ionic concentrations, which leads to calcium entrance and consequently the intracellular free calcium concentration rises. This increase in calcium at this point is extremely important for the formation of a new growth cone, as it triggers several cellular mechanisms fundamental for successful axon growth, namely: membrane sealing; local cytoskeleton transformation (typically actin and MT depolymerization) in order to form a growth cone; activation of local protein synthesis; activation of calpains (which proteolytically cleave submembrane spectrin) to allow vesicles fusion with plasma membrane; and activation of long-range retrograde molecular signalling. [56] This cytoskeleton remodelling is crucial to allow the transformation of a severed axon into a functional growth cone. Briefly, after calcium influx, MTs and actin depolymerize and calpain proteolitically digests submembraneous spectrin, allowing the plasma membrane to collapse. [63] Thereafter, there is an accumulation of vesicles in the ruptured site that will eventually fuse with the membrane and form a sealing patch. [64] After sealing, the excess intracellular calcium is removed, allowing for actin and MTs polymerization and therefore formation of a new growth cone. MTs repolymerize in random directions and anterogradely transported vesicles accumulate at their tips, whereas actin filaments assemble to generate the mechanical force at the leading edge of the lamellipodium. Of note, in the immediate moments after injury, the proteins needed come from recycling of the pre-existent material, while in later stages, the proteins needed in the reconstruction of the growth cone can either be

(36)

synthesized in the cell body and then anterogradely transported or locally synthesized. [56, 65]

In addition to calcium influx through the cut axonal end, axotomy also leads to calcium release from internal storages and to membrane depolarization, which activates voltage-gated calcium channels and further promotes calcium inflow. This calcium influx elicits a back-propagating calcium wave that reaches the soma and triggers epigenetic changes that will induce the expression of several genes, such as regeneration-associated genes (RAGs). [66] Interestingly, while PNS neurons can support long periods of high calcium concentrations and these are important for axonal regeneration, these are damaging for CNS neurons and therefore fail to upregulate RAGs, which may, in part, explain their lack of regeneration capacity. [51]

Furthermore, the early calcium influx also activates calcium dependent enzymes, such as adenylate cyclase, which leads to an increase in cAMP levels, which potentiate axonal regrowth. The increased levels of cAMP induce downstream dual leucine zipper kinase (DLK-1) expression, promoting local remodeling of the cytoskeleton needed for growth cone assembly. [67] Besides, when the calcium wave propagates into the cell body, protein kinase Cµ (PKCµ) is elicited and activates nuclear export of histone deacetylase 5 (HDAC5), which will increase histone acetylation and activate the pro-regenerative transcription program. Interestingly, in rodent sensory neurons, HDAC5 accumulates at the end of injured axons, where it promotes local tubulin deacetylation, inducing growth cone MT dynamics and therefore axon regeneration. [66] Although the calcium influx is critical for the establishment of the regeneration machinery, its duration and intensity depends on the neuron type and species. [56]

After the calcium dependent early phase, several injury signals, such as extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinases (JNK) and signal transducer and activator of transcription 3 (STAT3), are locally activated and retrogradely transported into the soma, more specifically to the nucleus. Some of these injury signals can activate a variety of transcription factors, for example: ERK activates ETS domain-containing protein (Elk-1), while JNK activates c-Jun and activating transcription factor 3 (ATF3). [68] Whereas JNK signalling has been implicated in the reorganization of the axonal cytoskeleton, STAT3 activation is important for neuronal survival after injury and both contribute to neuronal regeneration. Besides, the activated transcription factors can also induce the expression of several RAGs, like arginase 1, neuropeptide Y (NPY), vasointestinal peptide (VIP), interleukin-6 (IL-6), growth-associated protein-43 (GAP-43), among others. [51, 52] Some of the intracellular mediators involved in the post-injury response in PNS neurons are summarized in Figure 5. In summary, these proteins have

(37)

the capacity to alter the transcriptional profile of injured neurons in order to promote their survival and regeneration.

Figure 5 – Intrinsic mechanisms of axonal regeneration in PNS neurons after injury. After axonal severing,

target-derived negative signals are interrupted and repression of axonal elongation is relieved. Therefore, after calcium influx into the axoplasm, cAMP and PKA become activated, signalling to DLK-1, promoting membrane sealing, local protein synthesis and growth cone formation. When the calcium wave propagates to the soma, this leads to nuclear export of HDAC5 that together with retrogradely transported injury signals (such as ERK, JNK and STAT3) will activate t he proregenerative transcription program and promote expression of RAGs. RAGs (Arg1, NPY, VIP, IL-6, GAP-43, among others) are important to establish an effective regenerative response and after synthesis are then anterogradely transported. [51]

2.3. The Conditioning Lesion Model

Adult mammalian CNS neurons usually are not capable of regenerating. However, under certain conditions regeneration can occur, specifically by conditioning dorsal root ganglion (DRG) neurons, through an initial lesion of their peripheral axons that will allow regeneration of their central axons within the spinal cord. This is called the conditioning lesion effect. [69]

DRG neurons are a class of sensory neurons and present a pseudo-unipolar morphology with a peripheral axon branch, that is able to regenerate when injured, and a central axon branch that enters the spinal cord and does not regenerate upon injury. [52] As DRG neurons have a peripheral and a central branch, they have been extensively explored to study the regeneration process, given the possibility of evaluating differences in neuronal behaviour depending on the CNS and PNS environment. Therefore, in DRG neurons, if injury to the central branch is preceded by a lesion in the peripheral one, the central axon gains regenerative capacity and grows beyond the hostile lesion site. [70] This conditioning effect may be due to the previous activation of the regenerative machinery in that neuron, allowing to an effective and faster response after the second injury. In fact, it has been observed that there is an upregulation of RAGs and some proteins synthesized in the cell body, such as cytoskeletal proteins, are transported along

(38)

the regenerating axon where they may contribute to a reservoir of pre-made material that is at the disposal of the growth cone. Additionally, some proteins can also be synthesized near the tip of the axon, since specific mRNAs may be transported into the regenerating axon. [69, 71]

3. Profilin

Preliminary data from our group has demonstrated that the active form of profilin is increased in regenerating axons in vivo, which supports its relevance in axon formation and growth. Therefore, in this section, this actin-binding protein and its characteristics will be further described.

Profilin (Pfn) comprises a 14-17 kDa family of actin-binding proteins, expressed in all eukaryotes investigated [72] and in some viruses [73]. It was first described forty years ago, by Carlsson and colleagues [74] and so far four distinct profilin genes have been identified in mammals. [75] Profilin 1, whose gene is located in chromosome 17, is ubiquitously expressed, whereas the other forms have tissue-specific expression patterns. Profilin 2 (located in chromosome 3) has been reported to be alternatively spliced (in human, bovine, mouse and rat) into two isoforms, namely a major Pfn2A isoform, which is primarily expressed in neuronal cells and a minor Pfn2B isoform, which is mainly found in kidneys [76]. Pfn3 (chromosome 5) is kidney and testis specific, while Pfn4 (chromosome 2) is only testis specific, both playing a major role in acrosome formation and sperm morphogenesis. [77]

Curiously, despite having moderate amino acid sequence homology between the different profilin isoforms and also between profilins from distinct species, they arrange in quite similar three-dimensional structures. The ubiquitous distribution of profilin through eukaryotes, as well as the high level of conservation of the structure indicates an important role of profilin in cells [72, 78]. Typically, profilins display a compact centre of seven β-strands, surrounded by four α-helixes. Despite being a small protein, they have three main domains that allow interaction with other molecules, as depicted in Figure 6, namely: the actin-binding domain, responsible for its interaction with actin [79]; the phosphatidylinositol (PI) lipid-binding site, which allows interaction with PI-(4,5)-bisphosphate (PIP2) and other PI lipids [80]; and the poly-L-proline (PLP) binding

domain, which allows to interact with a plethora of proteins that contain PLP sequences [81]. Profilin contains two PI lipid-binding sites of which one overlaps the actin-binding domain and the other overlaps the PLP-binding site, meaning that interaction with PI lipids precludes interaction with either actin or PLP sequences. [82] These binding

(39)

domains are crucial for profilin functions and given the overlapping disposal of some of these domains, as mentioned above, this creates some regulatory complexity. Importantly, different profilin forms display distinct affinities to their ligands, for example, in humans, Pfn1 affinity for actin is higher than that for Pfn2, due to differential exposure of hydrophobic residues. [83]

Figure 6 – Profilin structure and its binding domains. Representation of human profilin 1 structure, showing actin

(orange), PIP2 (light brown) and PLP (blue) binding domains. [72]

3.1. The role of profilin in actin dynamics

As previously mentioned, profilin can interact with actin and therefore participate in the actin treadmilling. Profilin’s main function is to promote the transition of ADP-G-actin to ATP-G-ADP-G-actin, so that it can be incorporated in ADP-G-actin filaments and also mediates the release of actin monomers in F-actin barbed ends. Consequently, profilin is able to promote actin polymerization and increase actin dynamics. [42] Interestingly, profilin increases the nucleotide exchange rate in G-actin by 1000-fold, when compared with the rate based on simple diffusion, [84] which stresses its important role in actin dynamics and the associated cellular events. Importantly, it has been reported that the open nucleotide-site in G-actin is unstable in the absence of profilin, but is stabilized in its presence, promoting nucleotide exchange. [85] After the incorporation of G-actin in actin filaments, ATP is progressively hydrolysed by the intrinsic actin ATPase activity, generating ADP-actin in the older part of the filament, where other ABPs, such as cofilin, can operate, thus restarting a new cycle of actin polymerization. [42]

Given the above mentioned role of profilin in actin dynamics and the importance of actin dynamics in a plethora of cellular events, such as growth cone protrusion and axonal growth, it is expectable that its loss might have a pronounced impact in neurons.

(40)

Recent reports have demonstrated that Pfn1 depletion may cause significant defects in membrane extension and slower velocity of protrusion in both neuronal and non-neuronal migrating cells. [86, 87] This striking phenotypes can be explained by the fact that membrane protrusion depends on de novo actin nucleation or elongation of pre-existing filaments, which require the intervention of certain ABPs (Ena/VASP, WASP/WAVE, Arp2/3 complex and formins) that possess PLP regions. Besides profilin alone, some ABPs may also bind to profilin-G-actin-complex, enhancing F-actin elongation. [88] Therefore, this feature allows the interaction between these specific ABPs and profilin, further regulating actin dynamics. Specifically, Ena/VASP proteins promote elongation of actin filaments and tend to create unbranched and longer bundles of F-actin and by recruiting profilin-G-actin complexes, actin elongation is promoted. [89] Furthermore, the Arp2/3 complex, which can be recruited by WASP/WAVE proteins, promotes actin nucleation and may interact with profilin-G-actin complex to perform its function. [43] The same happens in formin-mediated actin nucleation, where profilin-G-complexes are also required. [90] Taken together, these reports support the view that profilin can interact concomitantly with actin and PLP regions, which further enriches the range of functions and resultant effects that Pfn may exert.

Besides the ability to interact with actin and PLP sequences, profilin can bind to PI lipids as well, being PIP2 one of best characterized PI lipids in terms of profilin

interactions and also the one with which it interacts with greater affinity. When profilin binds to PIP2, its hydrolysis by phospholipase C-γ (PLCγ) is inhibited. [91] However,

when PLCγ is phosphorylated in certain tyrosine residues, PIP2 is hydrolysed into second

messengers [92] and Pfn is released from the membrane to the cytosol, where it can interact with actin and other ligands. Therefore, given the PIP2 involvement in signalling

events, this suggests that Pfn may function as bridge between intracellular signalling and cytoskeleton remodelling.

It is also important to mention that profilin can be regulated by phosphorylation. Specifically, in the presence of PIP2, protein kinase C zeta (PKCζ) phosphorylates Pfn

at serine 137. [93] In addition, profilin may also be phosphorylated by Rho-associated protein kinase 1 (ROCK1). [94] This serine phosphorylation was reported to increase the affinity for G-actin and PLP stretches, while no effect was observed regarding its affinity to PIP2. [95] This increased actin affinity of phosphorylated profilin may result in

increased actin sequestering time and consequently lower actin incorporation in filaments.

(41)

3.2. Profilin in neuritogenesis

As aforementioned, during neuronal development, the symmetry of the initial cellular round spheres is broken, ensuing establishment of polarity and neuritogenesis, which greatly depend on actin dynamics. It has been previously proposed that both Pfn1 and Pfn2 participate in neuritogenesis. Specifically, it was observed that Pfn2a is a negative regulator of neuritogenesis, via ROCK, which is a downstream effector of RhoA. [96] When activated, ROCK may phosphorylate either Pfn1 or Pfn2 [94, 96], thus inactivating them and it may activate LIMK-1 which, in turn, inhibits cofilin [97], preventing actin severing. In fact, it was observed that an excess of intracellular Pfn2a blocks neurite formation and extension, whereas Pfn2a suppression does the opposite. [96] These effects seem to be due to increased and decreased actin-filament stability, respectively. Also, this implies that breaking the neuronal sphere requires a mechanism of inactivation of RhoA/ROCK/Pfn2 pathway, in order to reduce intracellular Pfn2 levels. A mechanism was proposed stating that, at early stages, there is a growth-positive stimuli capable of inactivating RhoA, which in turn dissociates from the membrane, reducing the interaction between ROCK and Pfn2a, avoiding Pfn2a phosphorylation. This translates into a localized shift in the local F/G-actin ratio, favouring the monomeric form and leading to actin instability and sprout formation. [96] It has also been reported that plasma membrane ganglioside sialidase (PMGS), which asymmetrically accumulates at the tip of the future axon in stage 2 neurons, can trigger PI3K- and Rac1-dependent inhibition of RhoA signalling. [98] Therefore, this would result in local actin instability and consequently axon specification.

Regarding Pfn1, it is known that it localizes at the leading edges of growth cones and it was proposed that this isoform also has a role in neuritogenesis. High-level overexpression of Pfn1 has been reported to inhibit neurite outgrowth, while low-level expression resulted in increased filopodia formation. [99] Interestingly, in a Pfn1 knockdown (KD) situation, studies revealed that there is a marked reduction of lamellipodia size, which were not able to efficiently extend. Therefore, Pfn1 KD leads to decreased actin dynamics and growth cone extension, [87] suggesting its importance in neuritogenesis. These differences between the role of Pfn1 and Pfn2 in neurite outgrowth could be explained, at least in part, by their different affinities for ligands.

Besides profilin function in the growth cone, it also has a significant role in regulating actin dynamics in dendritic spines. Dendritic spines are small actin-rich protrusions from neuronal dendrites that form the postsynaptic part of most excitatory synapses. Remarkably, modulation of actin dynamics drives morphological changes in

Exploring the role of Profilin 1 in axon formation, growth and regeneration